test: remove duplicate Pareto ordering probes / test: 删除重复的 Pareto 顺序测试

[functionstackx]

Summary

Delete five low-value tests from the paretoFront x-ordering for gradient labels block in packages/app/src/lib/chart-utils.test.ts.

57 test lines deleted; no production code, Pareto algorithm, UI/API behavior, browser spec, fixture, or timing-inventory changes. This is separate from the already merged InferenceTable cleanup in #1230.

Companion E2E cleanup: InferenceX #3574. Both PRs are independently mergeable.

Deleted test	Remaining proof
UpperRight ascending-order probe	The algorithm's handles unsorted input (sorts by x ascending internally) test asserts the complete ordered frontier.
LowerRight descending-order probe	Its handles unsorted input and ideal-staircase cases assert exact members and order.
LowerRight test-local ascending sort	This applies .sort() inside the test rather than exercising the chart's fix. Exact frontier tests remain, along with the gradient helper suite and rendered cost/energy gradient browser cases.
UpperLeft ascending-order probe	Its handles unsorted input case asserts the complete ordered result.
LowerLeft ascending-order loop	The supplied input produces one point, so the loop executes no assertions. The existing three-point unsorted-input and singleton-geometry cases remain.

No assertions need transferring for this batch. All algorithm-specific numerical, boundary, missing-data and rendering tests remain. The separate obsolete cost-filter candidate is not included.

Validation

• Baseline workspace summaries: app 6,009 passed / 4 skipped; constants 63 passed; DB 831 passed; MCP 25 passed; skills 351 passed. The command wrapper timed out after the summaries were written.
• Candidate app-wide unit run: 6,000 passed / 4 skipped / 4 timeout failures in three unchanged localization files, each at the existing five-second timeout.
• Isolated rerun of all three affected localization files plus chart-utils, Pareto labels and power curves: 311 passed with --maxWorkers=1, without changing any timeout or test.
• Candidate constants 63, DB 831, MCP 25 tests passed; typecheck, targeted oxlint/oxfmt and git diff --check passed.
• Local browser smoke was attempted against an E2E_FIXTURES=1 server but exceeded the 360-second command limit during integration tests. It is incomplete, not a pass. Full browser validation remains a CI gate.
• The retained owner browser spec, cypress/e2e/gradient-labels.cy.ts, subsequently passed all 16 tests against the same fixture server.
• Independent final preservation review: no P0–P2 findings and no unique lost behavior identified.
• OpenClaw Codex autoreview was attempted but returned reviewer_unavailable / engine_failed without a report. The independent review is a fallback, not a successful Codex run.

AI model disclosure

• GPT 6 Astra Fast: implementation, validation and PR preparation; runtime-provided identifier.
• Read-only discovery agent: candidate/keeper/production/history audit; exact model/version was not exposed and could not be verified.
• Independent preservation reviewer: final deletion review; exact model/version was not exposed and could not be verified.

PR creation only; no merge requested.

中文说明

删除 chart-utils.test.ts 中五个低价值测试，共减少 57 行。生产代码、Pareto 算法、UI/API 行为、浏览器测试、fixture 和分片耗时清单均未改动。本次与已合并的 #1230 表格测试清理独立。

对应的 E2E 清理见 InferenceX #3574，两个 PR 可独立合并。

三个顺序探测由各算法已有的精确结果测试覆盖；一个测试只验证在测试内部复制的 .sort()，没有执行图表中的修复；另一个 LowerLeft 测试实际只返回一个点，循环内没有执行任何断言。保留完整有序前沿、单点边界、数值行为、gradient helper 及 cost/energy 图表渲染测试，无需迁移断言。

基线各工作区测试汇总通过：app 6,009、constants 63、DB 831、MCP 25、skills 351，app 跳过四项；命令包装器在汇总写入后超时。修改后的 app 全量运行有 6,000 项通过、四项跳过，并在三个未修改的本地化文件中出现四个五秒超时。随后以单 worker 重跑这三个文件及三个图表相关文件，311 项全部通过，未放宽超时或修改测试。

修改后的 constants 63 项、DB 831 项、MCP 25 项通过，typecheck、目标文件 lint/format 和差异检查通过。已尝试本地浏览器 smoke，但 integration 阶段超过 360 秒限制，不能视为通过；完整浏览器验证仍交由 CI。

随后针对保留的 cypress/e2e/gradient-labels.cy.ts 运行浏览器验证，16 项全部通过。

独立覆盖保全审查未发现 P0–P2 问题或唯一行为覆盖缺失。Codex autoreview 返回 reviewer_unavailable / engine_failed，未生成报告；独立审查是替代审查。

AI 模型：GPT 6 Astra Fast 负责实现、验证与 PR 编写；只读发现 agent 和独立覆盖审查 agent 的精确模型名称或版本未由运行环境提供，无法核实。本 PR 保持未合并。

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
inferencemax-app	Ready	Preview	Sep 29, 2026 5:04am UTC